Determining the position of a dryer fabric by soft computing

Abstract

AB3:Paperikoneiden tuotantonopeuden kasvaessa automaatiojärjestelmien jainstrumentoinnin on kehityttävä vastaamaan haasteisiin. Yksi haasteista onpaperikoneen viiran paikan luotettava määrittäminen.

Modern paper machines produce paper at great speed and volume. The product can pass through the machine almost as fast as 1800 m/min. The paper machine thus sets high demands on its instrumentation. Part of this instrumentation is the equipment used for determining and controlling the position of the dryer fabric. -The measurement is based on measuring the amount of light emitted by diodes reflecting from the surface of a fabric. Due to the measurement conditions, the position is not linearly dependent on the signals generated by the measurement. To achieve reliable position information from the measured data, a combination of neural and fuzzy, i.e. soft computing is used. In this thesis a new method for this determination has been developed. -The role of fuzzy logic in an algorithm is to function as a pre-classifier. Such a classification is needed in order to direct the data to a suitable neural network. The valid data set is then analyzed by the neural network, which proposes the position of the fabric edge. Both the analysis of the network's training and the method which uses fuzzy logic and a neural network are presented in this work.